Magnetic coupling in epitaxial TM/MgO/Fe(001) (TM=FeCo, Fe/Co, Fe) macroscopic and microscopic trilayers

JOURNAL OF APPLIED PHYSICS VOLUME 94, NUMBER 6 15 SEPTEMBER 2003

Downl

Magnetic coupling in epitaxial TM ÕMgOÕFe„001… „TMÄFeCo, FeÕCo, Fe…macroscopic and microscopic trilayers

C. Martınez Boubetaa)

Instituto de Microelectrońica de Madrid-IMM (CNM-CSIC), Isaac Newton 8-PTM, 28760 Tres Cantos,Madrid, Spain

J. M. de TeresaFacultad de Ciencias, Instituto de Ciencia de Materiales de Aragoń, Universidad de Zaragoza-CSIC,50009 Zaragoza, Spain

J. L. Costa-Kramer and J. AnguitaInstituto de Microelectrońica de Madrid-IMM (CNM-CSIC), Isaac Newton 8-PTM, 28760 Tres Cantos,Madrid, Spain

D. Serrate, J. I. Arnaudas, and M. R. IbarraFacultad de Ciencias, Instituto de Ciencia de Materiales de Aragoń, Universidad de Zaragoza-CSIC,50009 Zaragoza, Spain

A. Cebollada and F. BrionesInstituto de Microelectrońica de Madrid-IMM (CNM-CSIC), Isaac Newton 8-PTM, 28760 Tres Cantos,Madrid, Spain

~Received 31 October 2002; accepted 11 June 2003!

Multilayered TM/MgO/Fe~001! heterostructures~TM: FeCo, Co/Fe, and Fe! are grown epitaxially,to study the dependence of the magnetic coupling between the two ferromagnetic electrodes on theinsulating MgO barrier width and the lateral dimensions of the structures. The crystal quality isinvestigated by reflection high-energy electron diffractionin situ at different growth stages of theTM/MgO/Fe~001! heterostructures. Magnetic characterization by superconducting quantuminterference device magnetometry~macroscopic structures! and transverse Kerr effect~microscopicstructures! shows clearly independent switching of top and bottom electrodes at large~above 20 Å!spacer thicknesses for macroscopic films. This independent switching is also observed on patternedstructures. For very thin barriers, decreasing the size of the elements in patterned arrays decreasesthe number of junctions coupled via pinholes. ©2003 American Institute of Physics.@DOI: 10.1063/1.1598280#

ucnn

p,u

inietere

tns

Tlec

ree

lec-lastifer-

de-thispin-uple

eticÅ,

ick-m-

thenholepy.as

cÅ

ith

INTRODUCTION

The fabrication of metal/insulator magnetic heterostrtures has special relevance in the field of magnetic tunjunctions ~MTJ!. These structures exhibit a field dependetunnel—or junction—magnetoresistance~TMR!,1 which hasapplications such as magnetic random access memories,gramable logic, field sensors, etc.2 In the MTJ research fieldmost of the fabricated structures, and especially those bwith transition-metal electrodes, are made of polycrystallferromagnets separated by amorphous insulating barrThese polycrystalline/amorphous systems are of great inest because of their performance and direct technologicalevance to industry. On the other hand, fully epitaxial~transi-tion metal/insulator! systems exhibiting TMR allow a direccomparison with state-of-the-art theoretical calculatiowhere Fe/MgO/Fe is among the most modeled systems.3

One of the crucial aspects in the performance of a Mis the presence of pinholes in the insulating barrier, reveato be very important in the transport behavior of tunnel juntions, but also in the magnetization reversal of multilayeheterostructures. In addition, pinholes can affect the indep

a!Electronic mail: [email protected]

4000021-8979/2003/94(6)/4006/7/$20.00

oaded 27 Aug 2010 to 161.111.180.103. Redistribution subject to AIP licen

-elt

ro-

ilters.r-l-

,

Jd-dn-

dent character of the magnetization reversal of both etrodes by direct ferromagnetic exchange coupling. Thiseffect on the hysteresis loops has been modeled for antromagnetically coupled multilayers,4 giving rise to ferromag-netic, antiferromagnetic, or to a mixture of both statespending on pinhole size and separation. However,characteristic is less studied, and neither the size of theholes nor their number or density needed to exchange cothe two electrodes are known.

In this sense, Keavneyet al.5 studied magnetic couplingin Fe/MgO/Fe epitaxial heterostructures; Fe ferromagnelectrodes were coupled for MgO thicknesses below 75and via transport measurements, no MgO spacers with thnesses below 40 Å were found to be insulating. The comented thickness dependence of the switching field wasassigned to a thickness dependence pinhole size or pindensity, as was observed by scanning electron microsco

Interlayer magnetic coupling through MgO spacers halso been studied by van der Heijdenet al.6 who fabricatedFe3O4 /MgO/Fe3O4 trilayers, again finding ferromagneticoupling due to pinholes for MgO thicknesses below 13and due to ‘‘orange peel’’ coupling7 effects above this thick-ness. Obviously, a tight control on the spacer quality, w

6 © 2003 American Institute of Physics

se or copyright; see http://jap.aip.org/about/rights_and_permissions

e

ind

nag.thn

ldtoc

c-tshendciv

—

Atiodeino-at

pet

ed

-la

ys,

a

lir

.

fof-te

sg-

edre,byof

reer-

erstopy

gOg-or-

onteby

ial

fect

e

ion

also

at-me

ac-the

xial

tentent

ten

Co

urO.

iesnt

the

4007J. Appl. Phys., Vol. 94, No. 6, 15 September 2003 Martinez Boubeta et al.

Downl

absence of pinholes is a big challenge, and was not until vrecently that Popovaet al. have shown their capability toobtain continuous insulating MgO spacers as thin as 8 Å inFe/MgO/Fe epitaxial heterostructures,8 and the observationof interlayer antiferromagnetic coupling mediated by sppolarized tunnel electrons between the two Fe electrothrough MgO layers as thin as 5 Å.9

That work claims agreement with theoretical predictioof antiferromagnetic exchange coupling between ferromnetic electrodes separated by ultrathin insulating spacers10,11

Finally, in the case of structures of lateral dimensions oforder of micrometers, dipolar interactions between top abottom electrodes are enhanced. This renders at zero fieantiparallel electrode alignment that minimizes the magnestatic energy of the system, as demonstrated in the specase of tiled Fe/MgO/Fe~001! trilayered structures.12

From the basic point of view, MTJs with identical eletrodes simplify the physical interpretation of the resulHowever, to observe TMR, an antiparallel orientation of telectrode magnetizations at selected field values is matory, and thus the two electrodes must have different coerfields. A different switching field can be achieved by~i!slightly varying the composition of the electrodes,~ii ! depo-sition of additional layers that antiferromagnetically pinexchange bias—the magnetization of one electrode, and~iii !coupling one electrode to additional ferromagnetic layers.an example of the first approach a study of the magnetizareversal in Fe0.5Co0.5/MgO/Fe(001) systems is describelater. The third approach, also reported in this work, involvmaking the top Fe layer magnetically harder by depositCo on top, which grows epitaxially as well. The two ferrmagnetic layers are exchange coupled and the magnetizreversal of the combined electrode occurs cooperatively.

Within this scope, this work presents results on the etaxy and magnetization behavior of metal/insulator magnheterostructures. The spacer is MgO and the electrodesFe, FeCo, and Fe/Co, which have been the electrodes usepitaxial magnetic tunnel junctions with MgO~001! spacersand demonstrated TMR.13,14 In addition, magnetization results on structures of both macroscopic and microscopiceral dimensions are shown.

EXPERIMENT

The structures are fabricated in an UHV deposition stem by the combined use of sputtering and laser ablationdescribed elsewhere,15 both on GaAs~001! or MgO~001! sub-strates. In both cases, a 100 Å MgO buffer layer is grown450 °C by normal incidence pulsed laser deposition frommonocrystalline MgO target in the pressure range of31029 mbar. Deposition rates around 0.17 Å/s were cabratedex situby x-ray low angle reflectometry, profilometeand transmission electron microscopy~TEM! measurementsThe role of this MgO buffer layer onto GaAs~001! is two-fold: it offers an appropriate symmetry and lattice matchthe epitaxy of bcc Fe~001! layers, and it acts as an interdifusion barrier to avoid incorporation of As from the substrainto the subsequent layers, showing excellent propertiediffusion barrier with thermal stability up to 800 °C and si


ry

es

s-

edan-

ific

.

a-e

sn

sg

ion

i-icare

in

t-

-as

ta7-

r

as

nificant good electrical insulation characteristics in layersystems.16 In addition, this homoepitaxial MgO buffer layeonto MgO~001! substrates flattens the original MgO surfaceliminating all defects and microscopic roughness createdexposure to atmospheric water vapor and the formationbrucite MgO(OH)2 .17

On this MgO buffer layer, Fe, Co, and FeCo layers adeposited from individual Fe and Co targets. This is pformed by triode sputtering at 431024 mbar Ar pressurewith deposition rates around 0.1–0.3 Å/s. MgO spacer laywere deposited at 731029 mbar by laser ablation and a400 °C substrate temperature. Auger electron spectroscmeasurements confirm the 1:1 composition for the Mbuffer layers and spacers18 as well as the absence of Fe seregation in the MgO and vice versa. This last point is croborated by TEM measurements.15 The crystalline structureis analyzed by reflection high-energy electron diffracti~RHEED! in situ. The surface morphology at intermediagrowth stages and for the final structure is also examinedatomic force microscopy~AFM!, scanning areas of up to 535 mm2. Magnetic properties are studied in a commercsuperconducting quantum interference device~SQUID! mag-netometer and by transverse magneto-optical Kerr ef~MOKE!. Finally, microfabrication techniques@photo andelectron~e!-beam lithography, plus Ar1 ion milling# are usedto define structures in the microscopic range.

RESULTS

The epitaxial growth of the different layers that form thentire structure is studied in detail by RHEEDin situ. Thegrowth of individual Fe, Co, and Fe0.5Co0.5 single layers isperformed to check their epitaxy previous to the depositof the multilayered structures. In Fig. 1~b! the RHEED pat-terns for a 200 Å Fe0.5Co0.5 film grown at RT are shown. TheRHEED patterns for the MgO buffer layer@Fig. 1~a!# arealso presented, and for the sake of comparison wepresent in Fig. 1~c! the RHEED patterns for an Fe filmgrown at RT and subsequently annealed at 400 °C. All pterns are recorded with the incident beam along the sacrystallographic directions. Notice sharp and intense diffrtion streaks with a well defined spacing that depends onin-plane crystalline direction. This demonstrates the epitanature of the film. The similarity of both Fe and Fe0.5Co0.5

RHEED patterns, added to the presence of 50% of Fe conin the alloy, allows the identification of a bcc crystallinstructure for the Fe0.5Co0.5 layer. This bcc phase is consistewith the bulk Fe–Co phase diagram.19 On the other hand, thelattice constant for Fe0.5Co0.5 is about 2.856 Å,20 which isclose to the 2.866 Å value for bcc Fe. Surface faceting, ofobserved in Fe grown on MgO at RT,15 is not present hereand even faint Kikuchi lines can be observed for the Fepattern aligned along the@100# direction. This confirms thehigh crystalline quality and improved surface flatness of oFeCo as-grown film compared to the RT Fe epitaxy on MgThis is probably due not only to different interface energbetween FeCo/MgO and Fe/MgO, but also to differeSchwoebel barriers distribution for surface diffusion thanones observed for Fe/Fe homoepitaxy.21


Å

is

ruC

thÅmnedac°

ex

n

ks

eTo

tepin

re-ers.byepre-

hethen byrre-ec--of

-and

eri-mp-m

-

gO/

ayer

4008 J. Appl. Phys., Vol. 94, No. 6, 15 September 2003 Martinez Boubeta et al.

Downl

On the other hand, the RHEED patterns for the 100Co film grown on Fe~001! are also shown in Fig. 1~d!, thesame observed symmetry and spacing with respect topointing to a bcc Co epitaxial growth at RT. This is constent with results of previous works22–24 where it has beendemonstrated that ultrathin Co films grown on Fe~001! up to35 ML display a metastable body-centered-tetragonal stture, the equilibrium hcp phase developing at largerthickness.

Figure 2 shows a sequence of RHEED patterns forindividual layers of a complete 100 Å Co/100 Å Fe/20MgO/400 Å Fe~001! epitaxial heterostructure. From bottoto top, first the pattern of the Fe bottom electrode is showFig. 2~a!. Deposition of Fe at RT gives rise to a facetsurface, as evidenced by the chevron like features. This fting disappears after annealing the Fe surface at 400which is going to be the deposition temperature of the nMgO layer. As shown in Fig. 2~b!, the Fe RHEED patternafter annealing exhibits sharp and intense streaks withindication of faceting. Deposition of MgO on this flat Fe filmyields sharp and intense RHEED diffraction streaas shown in Fig. 2~c!, confirming the epitaxialFe(001)@110#iMgO(001)@100# relationship. Finally, in Figs.2~d! and 2~e! the diffraction patterns corresponding to thsuccessive deposition of Fe and Co at RT are shown.first Fe film is faceted again, and the Co layer depositedtop is somehow more defective as compared to the patshown in Fig. 1~b!, probably due to the presence of hcgrains. The deposition of the upper electrode—and capp

FIG. 1. ~a!, ~b! RHEED patterns showing the epitaxy of FeCo~001! onMgO~001!. ~c! The pattern of a Fe~001! film on MgO~001! is also shown todemonstrate the equivalent symmetry.~d! RHEED pattern showing the epitaxy of Co on Fe~001!.


Fe-

c-o

e

in

e-C,t

o

,

henrn

g

layers—was performed at RT without any annealing, toduce as much as possible any interdiffusion between lay

We now focus on the magnetic characterizationSQUID of the Fe~Co!/MgO/Fe trilayers. As-grown squar535 mm2 samples were measured in these experiments,vious to any lithographic process. In Fig. 3~a!, the magneticloops at 20 and 300 K for Fe0.5Co0.5 ~200 Å!/MgO ~20 Å!/Fe~200 Å! are shown with the magnetic field applied along t@100# Fe direction. Notice first that, at both temperatures,magnetic electrodes are magnetically decoupled as showdistinct and rather sharp magnetization transitions cosponding to independent magnetic switching of the eltrodes. In Fig. 3~a!, the magnitude of the transition is compared with the expected value for antiparallel alignmentthe Fe bottom electrode and Fe0.5Co0.5 top electrode~dottedline! with the following assumptions:~1! the real thickness isthe nominal one,~2! the lower coercive field transition corresponds to the bottom electrode magnetization switch,~3! the magnetic moment for Fe and Fe0.5Co0.5 in the trilayeris the bulk value~2.22mB and 2.41mB per unit cell, respec-tively!. The agreement between this estimate and the expmental results is very good, which makes the earlier assutions plausible. Moreover, the coercive field of the bottoelectrode, Fe~200 Å!, in the trilayers at 300 K is about 10

FIG. 2. RHEED patterns showing the epitaxy of the complete Co/Fe/MFe~001! heterostructure.~a! Faceted Fe deposited at RT on MgO~001!, ~b!Fe RHEED pattern after heat treatment at 400 °C during 10 min,~c! 20 ÅMgO barrier, ~d! 100 Å Fe layer deposited at RT,~e! diffraction patterncorresponding to a deposit of 100 Å Co on the previous commented Fe lat RT.


ta

btulemtg

35y

w

ps

arsve

, ran

du

did

hi

ntmsre-re

odeb-erer-

forrestomvar-

er

Fe

O


Downl

Oe, which coincides with the value observed by CosKrameret al.25 for single Fe~200 Å! thin films. The fact thatthe coercive field of the bottom electrode is not influencedthe growth of the barrier and the top electrode is a signaof the absence of substantial coupling between both etrodes in this trilayer. The coercive field of this Fe bottoelectrode increases to about 16 Oe at 20 K, usually attributo the decrease of the thermal energy that assists the matization reversal mechanism~180° domain walls nucleationand propagation!. In the case of the top electrode Fe0.5Co0.5

~200 Å! the coercive field at 300 K is 24 Oe increasing toOe at 20 K. It is worth it to bear in mind that the FeCo allofor this composition displays a negativeK1 anisotropy con-stant, i.e., @110# crystallographic directions become nomagnetically easy and@100# hard.26

In Fig. 3~b!, a comparison of the magnetization looapplying the magnetic field along the@100# and @010# in-plane crystallographic directions of the bottom~Fe! electrodeis shown. For the field applied along the@100# direction, twowell-defined abrupt steps in the magnetization reversalobserved for each loop branch, corresponding to the reveof the top and bottom electrode magnetizations. These resals occur at moderate coercive fields of 16 and 35 Oespectively. On the other hand, on rotating the sample 90°applying the magnetic field along the other easy axis@010#Fe direction, an additional step at about 10 Oe is observethe magnetization reversal process. This has been previoobserved in single Fe~001! layers grown on MgO~001!substrates,25 and its origin has been attributed to an adtional uniaxial magnetic anisotropy of interfacial origin anprobably due to an in-plane distortion of the Fe lattice. T

FIG. 3. SQUID magnetization loops for a 200 Å FeCo/20 Å MgO/200 Åtrilayer. ~a! Temperature dependence of switching field at 20 and 300 K.~b!Evidence of uniaxial in plane anisotropy of the Fe bottom electrode.


-

yrec-

edne-

realr-

e-d

insly

-

s

additional uniaxial anisotropy renders magnetically differedirections that are cristallographically equivalent. This seeto be the case also for the FeCo/MgO/Fe trilayer. It ismarkable that for this high-quality crystalline heterostructuthe magnetic anisotropy properties of the bottom electr~closely related to its interfacial interaction with the sustrate! still manifest after deposition of the complete trilaystructure without being affected by the deposition of uppmost layers.27

Figure 4 shows the magnetic hysteresis loops at 20 Kthree 100 Å Co/100 Å Fe/MgO/400 Å Fe heterostructuwhere the composition and thickness of the top and botelectrodes remain unchanged and the MgO thickness isied from 40 Å@Fig. 4~a!# to 30 Å @Fig. 4~b!# and 20 Å@Fig.4~c!#. The hysteresis loop for the 40-Å-thick MgO spac

FIG. 4. SQUID magnetization loops at 20 K for 100 Å Co/100 Å Fe/Mg~d! /400 Å Fe heterostructures with varying MgO barrier thicknesses~a! d540 Å, ~b! d530 Å, and~c! d520 Å. Magnetic field was applied alongthe @100# Fe direction.


Otld-F

igza

o

erecrs

lsot

neygC

2ot

oic

agthei

Fere

-lthbsna

segiodic

cerleisdeÅ

orst

ran-es.

tedns inrKEthetheionon-ot-

opsng

rsalscro-

s a

Å


Downl

shows two clear magnetization reversals at 50 and 450approximately. This indicates totally independent magneswitching of top and bottom electrodes, with a large fierange~;300 Oe! with an antiparallel magnetization configuration. Taking into account the bulk magnetic moments of(2.22mB) and Co (1.72mB per unit cell! and using thenominal thickness of the electrodes, it is possible to assthe lower coercive field to the bottom electrode magnetition reversal and the higher coercive field to the top one@seeFig. 4~a!#. The measured high value of the coercive fieldthe top electrode, the bilayer Co/Fe~450 Oe!, shows thatdirect ferromagnetic coupling between Co and Fe is veffective increasing the coercive field of the combined eltrode. This effect has also been reported for Ni/Fe bilaye27

and by Faure-Vincentet al. for Co/Fe on MgO/Fe.9

The morphology of the individual electrodes was acharacterizedex situby AFM measurements for the differensamples. In the case of the bottom Fe layer, a rms roughof 3 Å was obtained. For the case of the double Co/Fe lastructure, AFM inspection revealed an increased rms rouness of 8 Å. This coarseness arises from the defectivedeposition on top the faceted Fe layer, as shown in Figs.~d!and 2~e!. Since orange-peel type of coupling between belectrodes would be ferromagnetic only for the caseroughness at the barrier/bottom electrode interface, whwould produce correlated interface topography of both mnetic layers in the case of MgO barrier conformal growth,orange-peel type of coupling in these heterostructures ispected to be negligible because of the sharp Fe/MgO/Feterfaces. Additional evidence of atomically flat Fe/MgO/interfaces comes by high-resolution TEM observationsported in Ref. 15.

For the thinnest MgO spacer~20 Å! the hysteresis loopis almost square@see Fig. 4~c!#, with a single abrupt magnetization reversal. This is due to the simultaneous reversatop and bottom layer magnetizations, probably due topresence of pinholes in the MgO spacer. The differencestween Figs. 3~a! and 4~c! for the same MgO barrier thicknescould indicate in this last case some deviation from nomideposited thicknesses.

In the case of the 30 Å MgO thickness@see Fig. 4~b!#,the situation is intermediate between the 20 and 40 Å caMagnetization reversal appears to proceed in three stawith two clear transitions at 50 and 400 Oe, and a rotatstage at intermediate fields characteristic of coupled meThis is probably because pinholes present in the 20-Å-thMgO spacer sample shown in Fig. 4~c! have almost com-pletely disappeared in the 30-Å-thick one.

Up to now, results on epitaxial trilayer systems of maroscopic~millimeter! dimensions have been presented, whdipolar type interactions due to edge accumulation of podo not play a relevant role. In what follows this studyextended to similar structures but of micrometer lateralmensions defined by photo and e-beam lithography. Kmagnetization loops for a nominal vertical structure of 100Co/100 Å Fe/30 Å MgO/400 Å Fe are shown in Fig. 5 felements of different shapes and sizes: circular element18, 28, and 43mm diameter@Fig. 5~a!# and a square elemenof 700 mm lateral dimensions@Fig. 5~b!#. All structures are


eic

e

n-

f

y-

sserh-o

hfh-

ex-n-

-

ofee-

l

s.es,nia.k

-es

i-rr

of

patterned on the same macroscopic sample, which guatees identical deposition conditions and layer thicknessWhile these circular structures can clearly be differentiafrom macroscopic samples, the square one has dimensiothe millimeter range. In Fig. 5~a! the hysteresis loops foeach circular trilayer structure measured by focused MOat RT are shown. The magnetic field is applied along@110# Fe hard axis. Even though the signal is noisy due tosmall amount of probed material, two clear magnetizatreversal steps are observed for the three junctions, demstrating independent magnetization switching of top and btom electrodes.

For the square shaped structure the Kerr hysteresis lo@Fig. 5~b!# measured with the magnetic field applied alothe in plane@100# easy and@110# hard Fe directions clearlyshow the two sharp and independent magnetization reveas well. These are similar to those observed for the mascopic samples shown in Fig. 4~a!. Despite of its still mac-roscopic size, the reduction in lateral dimensions produce

FIG. 5. MOKE hysteresis loops for a 100 Å Co/100 Å Fe/30 Å MgO/400Fe structure patterned into elements of different sizes and shapes.~a! Circu-lar elements of different sizes~magnetic field applied along@110# Fe direc-tion!, ~b! squared single element of 700mm lateral dimension.


arede

-e.he

bcencOinop

ivlFhr

g

rranisc

e,pthu

yti

wfoth

allrig5

sant

-an

a, tp

c-sre

iciedral

sti-f

er,

to

n ofromd


Downl

better definition of the magnetization reversal in this squstructure. In the case of the circular elements, and dutheir microscopic dimensions, an additional term due topolar coupling needs be taken into consideration to undstand their magnetization reversal,12 nevertheless, no significant differences are noticeable after comparison of hysterloops of circular junctions with those for square junctions

As pointed out before, for the thinnest MgO barriers thysteresis loop is nearly square due to the simultaneousversal of the top and bottom layer magnetizations, probaoriginated by the presence of pinholes in the MgO spaThe role of pinholes on the magnetization reversal for jutions as a function of the lateral size, for a thin 10 Å Mgbarrier, is further illustrated in Fig. 6. The film structurethis case is 100 Å Fe/10 Å MgO/100 Å Fe, i.e., identical tand bottom electrodes. As observed in Fig. 6~a!, the continu-ous macroscopic film displays a square loop with a coercfield of 10 Oe, demonstrating the simultaneous reversatop and bottom electrode magnetizations like in a singlelayer film.25 The sample is patterned by e-beam lithograpand ion beam etching, as described in Ref. 12, and thesulting array of square junctions is measured by focusinlight spot, about 100mm diameter, within the square 250mmedge array. The measured magneto optical Kerr loop cosponds then to the behavior of several hundreds to thousof microjunctions. Figure 6~b! shows selected hysteresloops for arrays with different lateral sizes of the microjuntions, with the field applied along an easy Fe@100# direction.For junctions larger than 3mm, the reduced remanencMR5M /MSuH50 , of the array is close to one, pointing to toand bottom Fe electrodes that are exchange coupled. In owords, most of the array still behaves like the continuofilm. However, for junctions smaller than 3mm the reducedremanence decreases, and loops evidence a reversal bstages with a remanent state close to zero magnetizaalong the applied field direction. This indicates that the telectrodes magnetizations orient in an antiparallel fashiona large fraction of the array elements, and is explained bymagnetostatic energy reduction obtained by the antiparorientation of the electrodes magnetizations. For this barthickness of 10 Å MgO, the effect of patterning on the manetic loops is different of what has been measured for 20,and 70 Å MgO,12 where the two stage magnetization reveris measured for all lateral sizes. Furthermore, and differefrom the case of single layer~200 Å Fe! tiles that break intodomains for sizes smaller 3mm,25 a more stable single domain structure due to the flux closure between the topbottom electrodes is achieved in each trilayered tile.12 Allthese experimental facts prove that pinholes in the barrierthe main cause of the observed behavior. In this schemesystem behaves as if there was a threshold number ofholes ~or pinholes of a certain size! above which the twoelectrodes are coupled ferromagnetically@see a simplified il-lustration in Fig. 6~c!#. In this scheme, patterning the struture and decreasing the size of the elements decreasenumber of junctions coupled via pinholes increasing the aoccupied by exchange uncoupled junctions.


etoi-r-

sis

re-lyr.-

eofeye-a

e-ds

-

ers

twoonoreeler-0,lly

d

rehein-

thea

CONCLUSIONS

Multilayered TM/MgO/Fe~001! heterostructures~TM:FeCo, Co/Fe, and Fe! are grown epitaxially and the magnetcoupling between top and bottom electrodes is studchanging the insulating MgO barrier thickness and the latedimensions of the structures. The crystal quality is invegated by RHEEDin situ at different stages of the growth o

FIG. 6. MOKE hysteresis loops for a 100 Å Fe/10 Å MgO/100 Å Fe trilay~field applied along the@100# Fe direction!. ~a! Continuous unpatterned film~b! arrays of squared junctions with different lateral sizes ranging from 92 mm, and different interelement separations.~c! simplified scheme of theeffect patterning has on the coupling mediated by pinholes as a functiothe trilayer lateral size. The percentage of tiles with pinholes is reduced f100%~1 out of 1! for the first case, 50%~1 out of 2! in the second case, an25% ~1 out of 4! in the third one.


r

t

picocithrioth

d.

gcia

ev

ys

L

lf

. M.

on-

nd


Downl

the TM/MgO/Fe~001! heterostructures. Magnetic characteization by SQUID ~macroscopic structures! and transverseKerr effect~microscopic structures! shows clear independenswitching of top and bottom electrodes at large~40 Å! spacerthickness. Remarkably, the bottom Fe electrode anisotroare unaffected by the subsequent layers depositions thatplete the heterostructure. The magnetic behavior on reduthe MgO epitaxial insulating barrier thickness indicatespresence of pinholes for the thinnest spacers. This givesto ferromagnetic coupling due to direct contact between belectrodes. However, for very thin barriers, patterningstructure and decreasing the lateral size of the elementscreases the percentage of junctions coupled via pinholes

ACKNOWLEDGMENTS

The Spanish Commission of Science and Technoloand Comunidad de Madrid are acknowledged for finansupport.

1J. S. Moodera, L. R. Kinder, T. M. Wong, and R. Meservey, Phys. RLett. 74, 3273~1995!.

2S. S. P. Parkinet al., J. Appl. Phys.85, 5828~1999!.3W. H. Butler, X. G. Zhang, T. C. Schultess, and J. M. MacLaren, PhRev. B63, 54416~2001!.

4J. F. Bobo, H. Kikuchi, O. Redon, E. Snoeck, M. Piecuch, and R.White, Phys. Rev. B60, 4131~1999!.

5D. J. Keavney, E. E. Fullerton, and S. D. Bader, J. Appl. Phys.81, 795~1997!.

6P. A. A. van der Heijden, P. J. H. Bloemen, J. M. Metselaar, R. M. Wo


-

esm-ngesethee-

y,l

.

.

.

,

J. M. Gaines, J. T. W. M. van Eemeren, P. J. van der Zaag, and W. Jde Jonge, Phys. Rev. B55, 11569~1997!.

7L. Neel, C. R. Hebd. Seances Acad. Sci.255, 1676~1962!.8E. Popovaet al., Appl. Phys. Lett.81, 1035~2002!.9J. Faure-Vincent J, C. Tiusan, C. Bellouard, E. Popova, M. Hehn, F. Mtaigne, and A. Schuhl, Phys. Rev. Lett.89, 107206~2002!.

10J. C. Slonczewski, Phys. Rev. B39, 6995~1989!.11P. Bruno, Phys. Rev. B52, 411 ~1995!.12J. L. Costa-Kra¨mer, J. Anguita, J. I. Martıń, C. Martınez-Boubeta, A.

Cebollada, and F. Briones, Nanotechnology13, 695 ~2002!.13M. Bowenet al., Appl. Phys. Lett.79, 1655~2001!.14E. Popovaet al., Appl. Phys. Lett.81, 1035~2002!.15C. Martınez Boubeta, E. Navarro, A. Cebollada, F. Briones, F. Peiro´, and

A. Cornet, J. Cryst. Growth226, 223 ~2001!.16W. H. Leeet al., Appl. Phys. Lett.77, 2192~2000!.17P. Liu, T. Kendelewicz, J. G. E. Brown, and G. A. Parks, Surf. Sci.412–

413, 287 ~1998!.18J. Palomares, C. Martıńez Boubeta, and A. Cebollada~unpublished!.19M. Hansen,Constitution of Binary Alloys~McGraw Hill, New York,

1958!.20W. C. Ellis and E. S. Greiner, Trans. Am. Soc. Met.29, 415 ~1941!.21K. Thurmer, R. Koch, M. Weber, and K. H. Rieder, Phys. Rev. Lett.75,

1767 ~1995!.22J. Zhang, Z. L. Han, S. Varma, and B. P. Tonner, Surf. Sci.298, 351

~1993!.23G. C. Gazzadi and S. Valeri, Europhys. Lett.45, 501 ~1999!.24S. K. Kim, C. Petersen, F. Jona, and P. M. Marcus, Phys. Rev. B54, 2184

~1996!.25J. L. Costa-Kra¨mer, J. L. Meneńdez, A. Cebollada, F. Briones, D. Garcıá,

and A. Hernando, J. Magn. Magn. Mater.210Õ1-3, 341 ~2000!.26Soshin Chikazumi,Physics of Ferromagnetism~Oxford University Press,

New York, 1997!.27M. Przybylski, J. Grabowski, F. Zavaliche, W. Wulfhekel, R. Scholz, a

J. Kirschner, J. Phys. D35, 1821~2002!.


https://www.researchgate.net/publication/235448018_Pinholes_in_antiferromagnetically_coupled_multilayers_Effects_on_hysteresis_loops_and_relation_to_biquadratic_exchange?el=1_x_8&enrichId=rgreq-b161e15fd2c630a0ea8003d317ed5921-XXX&enrichSource=Y292ZXJQYWdlOzIzNDkwNzc0MjtBUzoxMDEzMzQxMDc4MjAwMzNAMTQwMTE3MTM0MjMxNg==

https://www.researchgate.net/publication/235448018_Pinholes_in_antiferromagnetically_coupled_multilayers_Effects_on_hysteresis_loops_and_relation_to_biquadratic_exchange?el=1_x_8&enrichId=rgreq-b161e15fd2c630a0ea8003d317ed5921-XXX&enrichSource=Y292ZXJQYWdlOzIzNDkwNzc0MjtBUzoxMDEzMzQxMDc4MjAwMzNAMTQwMTE3MTM0MjMxNg==

Magnetic coupling in epitaxial TM/MgO/Fe(001) (TM=FeCo, Fe/Co, Fe) macroscopic and microscopic trilayers

Documents